In a press release published on the 26th of June, Cortex announced the launch of a deep learning and AI network for decentralised apps. As stated in the release this is the “first global decentralised deep learning and AI network”.
The network is especially designed for dApps that are powered by AI thus bringing deep learning models e.g. AI support to the blockchain ecosystem. This release comes after 15 months of hard work and is advertised as the first occurrence of artificial intelligence into a crypto network.
Creating the means for executing AI models on-chain, via a Graphics Processing Unit (GPU), in the Cortex Virtual Machine (CVM) was stated to be quite challenging for the Cortex team. On the other hand this effort unravels big possibilities to countless applications. DApp developers would incorporate trained AI models into smart contracts, but not before they are uploaded on the Cortex chain storage layer.
Ziqi Chen, Cortex CEO, commented on the capabilities of the Cortex Virtual Machine:
“On-chain machine learning is an extremely complex endeavor due to the computational demands, and the need to create a virtual machine that is Ethereum Virtual Machine compatible. With the Cortex Virtual Machine, we’ve achieved a breakthrough that brings the benefits of artificial intelligence to a wider audience. Although dApp developers will be among the first beneficiaries of the Cortex mainnet, this is only the beginning. In time, we expect to develop a diverse range of use cases, all delivered on-chain,”
Cryptocurrency lending, anti-fraud reporting, P2P financing platforms, insurance or creating credit reports for the DeFi industry are some of these potential use cases. Ziqi Chen believes that eSports and the gaming fields can also take advantage from a “diverse range of use cases”. He added that:
“In the near future, we expect to see stablecoins based on machine learning, decentralized decision making, malicious behavior detection, smart resource allocation, and much more. These are challenges that all intersect with crypto networks, where having trained AI models that are accessible on-chain will prove to be extremely valuable.”
At the moment, the Cortex mainnet includes 23 AI models that are trained with four datasets. The CVM is backward-compatible with the EVM and is able to run traditional smart contracts in addition to AI smart contracts.
The Cortex roadmap includes plans for mutual work with developers in implementing AI dApps on the network and providing on-chain machine learning to networks outside of Ethereum as well.